Fisher-Scale Variance Propagation for Artifact-Corrected Correlations

Psychometric meta-analysis frequently corrects observed correlations for statistical artifacts such as unreliability and direct range restriction prior to synthesis. Prior work has already developed sampling-variance formulas for corrected correlations, most notably in the validity-generalization literature. However, those developments were framed primarily for corrected-correlation estimation rather than for Fisher-scale random-effects meta-regression. This note presents a Fisher-scale variance-propagation approach for artifact-corrected correlations. The method treats the corrected correlation as a deterministic function of the observed correlation, propagates sampling variance through that correction map using the delta method, and expresses the resulting study-level variance on the Fisher scale used for analysis. The estimator nests the standard Fisher variance when no correction is applied and can be used directly in conventional REML meta-regression. Preliminary simulations indicate that Fisher-scale variance propagation improves calibration of moderator inference relative to fixed-correlation approximations under heterogeneity and unequal study weights. This note is released to establish priority for this Fisher-scale reformulation prior to a fuller theoretical and empirical treatment.